Journal of the Torrey Botanical Society 137(2–3), 2010, pp. 133–156

Phylogenetic analyses of morphological and molecular data reveal major clades within the perennial, endemic western North American Apiaceae subfamily Apioideae1 Feng-Jie Sun2 and Stephen R. Downie3 Department of Plant Biology, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801 USA

SUN, F.-J. AND S. R. DOWNIE (Department of Plant Biology, 265 Morrill Hall, 505 South Goodwin Avenue, University of Illinois at Urbana-Champaign, Urbana, IL 61801 USA). Phylogenetic analyses of morphological and molecular data reveal major clades within the perennial, endemic western North American Apiaceae subfamily Apioideae. J. Torrey Bot. Soc. 137: 133–156, 2010.—The taxonomy and phylogeny of the perennial North American Apiaceae subfamily Apioideae endemic to western North America (north of Mexico) have posed great challenges to systematists. Available classifications based on morphological characters are in general inconsistent and unsatisfactory, and cladistic analyses based on these data are limited to only a few taxa and a small number of characters. In this study, we scored 54 morphological characters from 123 taxa of North American Apioideae (representing 111 species in 21 genera) to construct an estimate of phylogenetic relationships and to compare the results obtained with those inferred for the group through previous studies using molecular data. The morphological and combined (morphological and molecular) datasets were analyzed using maximum parsimony (with equal, proportional, and successive approximations weighting strategies and Goloboff fit criterion applied to the morphological characters) and Bayesian approaches. Phylogenetic trees derived from morphological characters are largely congruent with those derived from molecular data, upon the collapse of weakly supported branches. The least number of most parsimonious trees is derived from the combined analysis when morphological characters are given proportional weights, and these trees are fully congruent with those derived from molecular data alone. The results revealed that many morphological characters used previously to delimit genera are highly homoplastic, such as the presence of a carpophore, stylopodium, pseudoscape, and dorsal wings, the number of vittae, and the orientation of fruit compression. The results also supported the monophyly of the group, in accordance with previous molecular studies. Three major clades and several well- supported subclades are tentatively circumscribed, thus facilitating future phylogenetic and revisionary studies. Key words: Aletes, cpDNA, Cymopterus, Glehnia,ITS,Lomatium, Podistera, Pseudocymopterus, Pteryxia, Zizia.

As one of the major centers of geographical 1 This work was supported by NSF grants DEB distribution of Apiaceae, western North 9407712 and DEB 0089452 to S. R. Downie, and by America (specifically, Pacific North America Francis M. and Harlie M. Clark research support grants, a Herbert Holdsworth Ross Memorial Fund and the adjacent Rocky Mountains, north of award, a Program in Ecology and Evolutionary Mexico) hosts some 200 species of Apiaceae Biology Summer Research Fellowship, and travel subfamily Apioideae (Mathias 1965). The grants from the Graduate College, the John R. taxonomy of these plants has been investigated Laughnan Fund, and the Program in Ecology and Evolutionary Biology to F.-J. Sun. for over a century, many different classifica- 2 Author for correspondence. E-mail: fsun@ tion systems have been proposed (e.g., Torrey illinois.edu. Current address: W.M. Keck Center and Gray 1840; Coulter and Rose 1900; for Comparative and Functional Genomics, Roy J. Mathias 1930; Mathias and Constance 1944– Carver Biotechnology Center, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801 1945; Cronquist 1997; S. Goodrich et al., USA. unpublished data), and new species and 3 The authors thank Ronald L. Hartman for combinations from the region are continuous- providing plant material and facilitating collecting ly being described (e.g., Hartman 1985, 1986, trips in the Rocky Mountains, Geoffrey A. Levin for help with the phylogenetic analyses, Deborah S. 2000, 2006; Hartman and Constance 1985, Katz-Downie for assistance in the laboratory, the 1988; Kagan 1986; Hartman and Kirkpatrick curators of the herbaria cited in the text for access to 1986). Recently, however, molecular phyloge- specimens, and two anonymous reviewers for netic studies based on DNA sequence data comments. This paper represents a portion of a have greatly challenged the morphology-based Ph.D. thesis submitted by F.-J. Sun to the Graduate College of the University of Illinois at Urbana- classifications of these taxa. While these Champaign. molecular studies supported the monophyly

133 134 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137 of this group of perennial, endemic taxa, they distribution, representing 21 genera, 111 spe- also revealed that most of the genera compris- cies, 10 varieties, and two subspecies, was ing the group are not monophyletic (Downie examined (Appendix). The ranges of several et al. 2002; Sun 2003; Sun et al. 2004; Sun and species reach into central NA; a few others Downie 2004; Sun and Downie, 2010). As extend into eastern NA, or are restricted to examples, Cymopterus Raf. and Lomatium that region. These taxa represent the same Raf., the two largest genera within the group accessions as examined previously for nrDNA and representing over half of all of its included ITS (Downie et al. 2002; Sun et al. 2004) and species (Kartesz 1994), are each highly poly- cpDNA rps16 intron (Sun and Downie 2004) phyletic, with species from each genus allying and trnF-trnL-trnT (hereafter trnF-L-T; Sun closely with many other genera. Available and Downie, 2010) sequence variation. For classifications of the group based on morpho- ease of comparison with the results of our logical characters are in general inconsistent earlier studies of the group, the nomenclature and unsatisfactory, and previous cladistic of the Cymopterus acaulis and Pteryxia studies based on morphology are limited to terebinthina species complexes are maintained only a few taxa and a small number of as in Kartesz (1994). Cymopterus glomeratus characters (Gilmartin and Simmons 1987; (Nutt.) DC. (5C. acaulis Raf.) traditionally Downie et al. 2002). Many taxa demonstrate has five infraspecific taxa, but on the basis of overlapping patterns of morphological char- the results of numerical multivariate analyses, acter variation, both at the intraspecific and we proposed that plants in this species interspecific levels (Mathias 1930; Hartman complex be recognized as a single species, 1985; Hartman and Constance 1985; Sun et al. with no varieties (Sun et al. 2005). Similarly, 2005, 2006, 2008), and morphological synapo- four varieties were recognized previously in morphies useful to circumscribe genera or Pteryxia terebinthina, but results of our prior major clades are few or heretofore unknown. multivariate analysis of this complex support- In this study, we use a phylogenetic ed only two, vars. foeniculacea and tere- approach to examine the morphological char- binthina (Sun et al. 2008). Based on results of acters used previously to circumscribe genera previous molecular studies, Aethusa cynapium within the perennial, endemic western North L. was chosen in the phylogenetic analyses to American (NA) Apiaceae subfamily Apioi- root all trees. deae. The major objectives of this study are to: Microscope slides of mature fruit cross- (1) construct an estimate of phylogenetic sections were prepared from two or more relationships within the group using morpho- herbarium specimens for nearly all species logical data; (2) evaluate the utility of mor- examined in this study. Prior to sectioning, phological data in circumscribing genera and fruits were softened by treating them for one major clades inferred on the basis of combined to two hours in warm water. Free hand morphological and molecular evidence; and sections through the middle of the mature (3) assess patterns in the evolution of several mericarps were made using a razor blade. specific morphological characters that have These sections were examined under an been widely utilized in previous classifications Olympus compound microscope for orienta- of the group (i.e., the development of a tion of fruit and seed compression, features of pseudoscape and a stylopodium, the pattern the ribs, wings, and commissure, and the of fruit compression, the development of a number and position of vittae. A total of 54 carpophore and fruit ribs, and the number of characters was scored; 26 of these were vittae in each interval of the fruit). Based on obtained from fruits, 14 from inflorescences, the combined morphological and molecular 11 from plant habit, and three from flowers. evidence, the monophyly of this group of These characters and their character states are perennial, endemic NA genera can be further provided in Table 1, along with additional evaluated and its major clades be circum- comments. The data matrix is presented in the scribed, thereby facilitating future phylogenet- Appendix. For the majority of these morpho- ic and revisionary studies. logical characters, the determination of their character states was obvious due to their Materials and Methods. ACCESSIONS AND qualitative nature. For nine quantitative char- MORPHOLOGICAL CHARACTERS EXAMINED.A acters (Table 1; Nos. 11, 17, 18, 31, 32, 35, 45, total of 123 taxa of primarily western NA 47, and 50), character states were determined 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 135

Table 1. Morphological characters and character states used in the phylogenetic analyses of 123 representatives of NA Apioideae.

Characters Character states and comments 1. Plant habit 0 5 acaulescent; 1 5 caulescent 2. Herbage habit 0 5 mat-forming; 1 5 stem one-few, tufted 3. Herbage surface 0 5 glabrous; 1 5 pubescent; 2 5 scabrous or granular 4. Root habit 0 5 tap, slender or thickened; 1 5 tap, tuberous or globose; 2 5 fibrous, fascicled 5. Root habit 0 5 branching caudex; 1 5 simple, not branching root crown 6. Peduncle surface 0 5 glabrous; 1 5 pubescent; 2 5 hirtellous or scabrous at summit 7. Pseudoscape 0 5 present; 1 5 absent 8. Leaf complexity 0 5 ternate pinnate once; 1 5 ternate-pinnately two to several times; 2 5 simply pinnate or subbipinnate 9. Leaf margin 0 5 irregularly toothed; 1 5 evenly serrate or dentate; 2 5 entire 10. Sheath 0 5 not or slightly ampliate; 1 5 conspicuously sheathing 11. Ratio of ultimate leaf segment 0 5,10; 1 5.10. Two patterns of ratio values were length / width observed: one group having most species with ratio values less than five (occasionally about six to eight), and the other group having ratio values larger than 10 (range 10–25). Therefore, the ratio value of 10 was used as a gap to distinguish these two characters states. One exception was found in Pseudocymopterus montanus, which has a polymorphism recorded, because no clear gap was found in the ratio values (range one to 15) in this species. This is likely due to the fact that P. montanus is such a variable species with regard to its leaf morphology. 12. Flower petal color 0 5 white; 1 5 purple or pinkish; 2 5 yellow; 3 5 green 13. Flower anther color 0 5 purple; 1 5 yellow; 2 5 white 14. Pedicels of sterile flowers 0 5 rigid and persistent; 1 5 neither rigid nor persistent 15. Inflorescence habit 0 5 spreading; 1 5 compact; 2 5 globose head 16. Primary ray surface 0 5 glabrous; 1 5 pubescent; 2 5 hirtellous or scabrous at summit 17. Primary ray length 0 5 equal or nearly equal (ratio of shortest ray length / longest ray length . 0.8); 1 5 unequal (ratio of shortest ray length / longest ray length , 0.8). Most species having state 1 were found to have ratio values less than 0.5. Others had ratio values . 0.8. The ratio value of 0.8 was used as a gap to distinguish the two character states. Two species, Polytaenia texana and Pseudocymopterus montanus, were recorded as having polymorphisms. Both species were found to be variable in this character. 18. Maximum primary ray number 0 5 less than 30; 1 5 more than 30. Two patterns of maximum primary ray number were found, with one group having most species with the maximum primary ray number less than 10 (several less than 20–25), and the other group having a maximum primary ray number . 30 (range 30–50). Therefore, the number 30 was used as a gap to distinguish these two character states. 19. Bract 0 5 present; 1 5 absent 20. Bract texture 0 5 entire herbaceous or with narrow scarious margin; 1 5 mostly scarious except midvein 21. Bract top edge 0 5 entire and tapering; 1 5 more or less obovate; 2 5 toothed 22. Involucre shape 0 5 nearly complete, forming a cup underneath umbel; 1 5 individual lobes, not forming a cup 23. Bractlet 0 5 present; 1 5 absent 24. Bractlet top edge 0 5 entire and tapering; 1 5 more or less obovate; 2 5 toothed 25. No. of midveins on the bractlet 0 5 one; 1 5 more than one 26. Bractlet color 0 5 white; 1 5 purple or green 27. Bractlet texture 0 5 entirely herbaceous; 1 5 herbaceous with narrow scarious margin; 2 5 mostly scarious except midvein 28. Involucel shape 0 5 nearly complete, forming a cup underneath umbellet; 1 5 dimidiate, not forming a cup 136 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137

Table 1. Continued.

Characters Character states and comments 29. Ovary surface 0 5 glabrous;1 5 pubescent or villous; 2 5 scabrous or granular 30. Stylopodium in fruit 0 5 absent; 1 5 present 31. Style orientation 0 5 widely spreading (angle between two styles . 45 degrees); 1 5 more or less erect (angle between two styles , 45 degrees). The angle between two styles was larger than 90 degrees for species having state 0, while smaller than 30 degrees for species having state 1. Therefore, the clear gap of 45 degrees was chosen to separate these two states. 32. Calyx teeth in fruit 0 5.0.6 mm, well-developed; 1 5,0.6 mm, not well- developed. Using calyx teeth length 0.6 mm as a gap to distinguish two character states was based on the fact that species having state 1 always had calyx teeth shorter than 0.5 mm and species having state 0 always had calyx teeth longer than 1 mm. 33. Fruit attachment 0 5 sessile; 1 5 pedicellate 34. Fruit surface 0 5 glabrous; 1 5 pubescent or villous; 2 5 scabrous or granular 35. Fruit compression 0 5 dorsally compressed (ratio of length of commissural face / width of two mericarps . 1.5); 1 5 laterally compressed (ratio of length of commissural face / width of two mericarps , 0.6); 2 5 terete (ratio of length of commissural face / width of two mericarps 5 0.9–1.1). Three groups of ratio values (, 0.6, 0.9–1.1, and . 1.5) were found and used to distinguish three characters states in this continuous character. 36. Carpophore 0 5 persistent; 1 5 present, but falling with mericarp; 2 5 absent 37. Carpophore branching 0 5 entire, not bifid; 1 5 bifid 38. Commissure 0 5 constricted (constricted . 80% of the commissural face); 1 5 not constricted (not constricted , 20% on the commissural face) 39. Corky and rib-like projection of 0 5 present; 1 5 absent fruit axis 40. Fruit ribs 0 5 all ribs winged; 1 5 only lateral ribs winged; 2 5 no wings 41. Fruit ribs 0 5 filiform; 1 5 rounded, corky 42. Fruit wings 0 5 chartaceous; 1 5 thick, corky 43. Dorsal wings 0 5 wavy or corrugated; 1 5 not wavy or corrugated 44. Lateral wings 0 5 wavy or corrugated; 1 5 not wavy or corrugated 45. Lateral wings 0 5 wider than fruit body (ratio of lateral wing length on cross- section / fruit body length . 1.2); 1 5 equal to fruit body (ratio of lateral wing length on cross-section / fruit body length 5 0.9–1.1); 2 5 narrower than fruit body ratio of lateral wing length on cross-section / fruit body length , 0.8). Most species having state 0 had ratio values larger than 1.5 (a few cases 1.2–1.5) and most species having state 2 always had ratio values smaller than 0.5 (a few cases 0.6– 0.8). The majority of species having state 1 had ratio values always about one. Clear gaps in this continuous character were used to distinguish these three character states. 46. Lateral wing 0 5 incurved (lateral wing is almost perpendicular to commissural face); 1 5 not incurved (lateral wing is parallel to the commissural face) 47. Seed compression 0 5 dorsally compressed (ratio of seed length on commissural face / width of one mericarp . 1.5); 1 5 laterally compressed (ratio of seed length on commissural face / width of one mericarp , 0.6); 2 5 terete (ratio of seed length on commissural face / width of one mericarp 5 0.9– 1.1). Three groups of ratio values (, 0.6, 0.9–1.1, and . 1.5) were found in this continuous character and used to distinguish these three characters states. 48. Seed face in cross-section 0 5 plane; 1 5 concave (at least halfway concave into seed) 49. Wing on cross-section 0 5 base enlarged; 1 5 not enlarged; 2 5 top enlarged 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 137

Table 1. Continued.

Characters Character states and comments 50. Ratio of wing length / wing width on 0 5,5; 1 5.5. Most species having state 0 had the ratio cross-section values about one to three, whereas species having state 1 had the ratio values larger than six. Therefore, the ratio value of five was selected as a gap to distinguish these two character states. 51. Strengthening cells 0 5 present; 1 5 absent 52. No. of oil tubes in the interval 0 5 one; 1 5 more than one; 2 5 inconspicuous; 3 5 none. State 2 indicates that the boundaries of the oil tubes are inconspicuous, such that their numbers cannot be counted. 53. No. of oil tubes in the commissure 0 5 two; 1 5 more than two; 2 5 inconspicuous. State 2 indicates that the boundaries of the oil tubes are inconspicuous, such that their numbers cannot be counted. 54. Accessory oil tube in rib 0 5 present; 1 5 absent

by detecting gaps in the character variation strap (BS) values (Felsenstein 1985) were (Stevens 1991). Character polymorphism and calculated from 100,000 replicate analyses uncertainties were observed and specified in using ‘‘fast’’ stepwise addition of taxa; only the data matrix (Appendix). Approximately those values compatible with the majority rule 10% of the cells in the data matrix were scored consensus tree were recorded. The number of as unknown or inapplicable. additional steps required to force particular taxa into a monophyletic group was examined PHYLOGENETIC ANALYSIS. The matrix of using the constraint option of PAUP*. The morphological characters was first analyzed pattern of evolution of each morphological using maximum parsimony (MP), with the character across one arbitrarily selected min- character state changes either equally or imal length tree was assessed using MacClade proportionally weighted. The latter was done vers. 4.0 (Maddison and Maddison 2003), with because the number of states differed among the goal of finding those characters most characters (ranging from two to four), so all useful for delimiting clades and, ideally, characters were weighted in inverse proportion genera. MacClade’s trace character or chart to their minimum number of steps using the option was used to determine the number of scale option of PAUP* (Swofford 2003). steps of each character over a randomly Another character weighting approach, suc- chosen tree or all MP trees. cessive approximations (Farris 1969), was also Bayesian analysis running four million used. Here, two successive weighting searches generations was carried out using MrBayes were done, one starting with equal weights and vers. 3.0 (Ronquist and Huelsenbeck 2003), the other with proportional weights, and the with tree sampling occurring every 100 gener- results from both searches were compared. In ations. This was done using the standard this approach, characters were weighted by the model for unordered characters with a stan- maximum values of their rescaled consistency dard gamma distribution to accommodate the (RC) indices, and searches were ended when rate variation across sites. Starting trees were the RC values became stable for at least three chosen at random and four simultaneous iterations. The matrix was also analyzed using Markov chain Monte Carlo chains were used equally weighted MP with Goloboff fit crite- to model the character rate heterogeneity. The rion selected (Goloboff 1993; K 5 2, default in posterior probability (PP) values (expressed as PAUP*). All character states were assumed percentages) for each bipartition of the phy- unordered, and the options multrees, collapse, logeny were determined from the remaining and acctran optimization were chosen. Due to trees after the removal of ‘‘burn-in’’ trees. the large number of taxa, MP trees were By using a ‘‘total evidence’’ analysis (Kluge sought using the heuristic search strategies of 1989; Kluge and Wolf 1993), also called a PAUP* and the inverse constraint approach ‘‘simultaneous analysis’’ (Nixon and Carpen- described in Catala´n et al. (1997) and later ter 1996), both molecular (ITS, rps16 intron, implemented by Downie et al. (1998). Boot- and trnF-L-T) and morphological data were 138 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137 combined into a single matrix for simulta- phological characters are congruent with those neous consideration. For each taxon with derived from molecular data from our previ- multiple accessions in the molecular datasets ous studies. All morphological analyses sup- (i.e., two accessions each of Aletes acaulis, port the monophyly of the group of perennial, Pseudocymopterus montanus, Pteryxia tere- endemic NA taxa. binthina var. albiflora,andPteryxia tere- binthina var. calcarea, and three accessions of MORPHOLOGY:MAXIMUM PARSIMONY USING Aletes macdougalii subsp. breviradiatus), the PROPORTIONAL WEIGHTS. MP analysis of 54 same morphological character states were morphological characters using proportional assigned to each taxon based on an examina- weights (i.e., 31 characters with a weight of tion of their voucher specimens. Therefore, the 1.00, 21 characters with a weight of 0.50, and final combined dataset contained 129 taxa. two characters with a weight of 0.33) resulted The protocols for searching for the most in the preset limit of 20,000 most parsimonious parsimonious trees using MP are the same as trees, each of 331.83 steps [consistency index those performed for morphological data. The (CI) 5 0.16; retention index (RI) 5 0.68; RC successive approximations (Farris 1969) and 5 0.11]. By using the inverse constraint the MP with Goloboff fit criterion selected approach, the strict consensus of the 20,000 (Goloboff 1993) were not performed. In the most parsimonious trees served as a topolog- Bayesian analysis, different models of maxi- ical constraint in a further heuristic search. In mum likelihood were given to different mo- this search, five more trees of the same length lecular partitions (ITS, rps16 intron, trnF-L-T) as these 20,000 trees were obtained. The strict of the combined data, as described previously consensus tree of these 20,005 trees is given in (Sun 2003; Sun et al. 2004, Sun and Downie Fig. 1. On this tree, Angelica capitellata is 2004; Sun and Downie, 2010). sister to the perennial endemic NA genera group, the latter comprising a large polytomy,

Results. DATA MATRICES. The combination and a clade of all remaining Angelica species is of molecular and morphological data for 129 successively basal to the aforementioned taxa. taxa resulted in a matrix of 3586 (3532 Coming off this polytomy, eight branches molecular, 54 morphological) characters, with contain four or more taxa, but the BS values no positions excluded from the molecular for all of these branches are low (,50%). partition because of alignment ambiguity. Among the 54 characters examined, six occur The combined dataset had a total of 408 without homoplasy (CI 5 1.00) on one parsimony informative characters (354 molec- arbitrarily selected MP tree (Nos. 14, 20–22, ular, 54 morphological). The values of the g1 39, and 46; Table 1). Of those genera tradi- statistics for 10,000 and 100,000 random trees tionally recognized within the group, only two of both morphological (20.148 to 20.172) are monophyletic (Oreonana and Orogenia), and combined (20.235 to 20.276) datasets and these genera are supported by non- were significantly more skewed than random homoplastic characters. Rigid and persistent data (20.09 to 20.11, P , 0.01), indicating pedicels of sterile flowers (No. 14, state 0) that these data contain significant amounts of support the clade of Oreonana, and the phylogenetic signal (Hillis and Huelsenbeck presence of corky and rib-like projections on 1992). fruit axes (No. 39, state 0) and incurved lateral wings (No. 46, state 0) support the clade of MORPHOLOGY. The results of all phylogenet- Orogenia. Characters having high CI values ic analyses of morphological data showed (.0.70) but show some homoplasy include similar results, differing primarily in their plant habit (No. 1; CI 5 0.87), flower petal degree of resolution and branch support. To color (No. 12; CI 5 0.72), and shape of the top show these relationships, we present only the of the bractlet (No. 24; CI 5 0.82). Many results of the MP analysis using proportional characters emphasized in previous classifica- weights because the strict consensus tree tion systems show high levels of homoplasy, resulting from this analysis (Fig. 1) showed such as the presence/absence of a pseudoscape greatest resolution and branch support over- (No. 7; CI 5 0.53), fruit compression patterns all. In general, upon the collapse of weakly (No. 35; CI 5 0.14), development of a supported branches (i.e., BS or PP values carpophore (No. 36; CI 5 0.43), development ,50%), phylogenetic trees derived from mor- of fruit wings (No. 40; CI 5 0.35), develop- 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 139

FIG. 1. Strict consensus tree of 20,005 minimal length trees derived from proportionally weighted MP analysis of 54 morphological characters from 123 members of NA Apioideae. Numbers on branches are bootstrap estimates (BS) for 100,000 replicate analyses using ‘‘fast’’ stepwise addition and Bayesian posterior probability (PP) values expressed as percentages, respectively; values ,50% for both support values are not indicated or indicated by ‘‘--’’. 140 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137 ment of a stylopodium in fruit (No. 30, CI 5 is the sister taxon of the NA genera group and 0.50), and number of oil tubes in the interval Angelica (excluding A. capitellata) is placed and on the commissure of the fruit (Nos. 52 one node away. Angelica capitellata is sister to and 53; CIs 5 0.26 and 0.23, respectively). a clade comprising all aforementioned taxa. Characters exhibiting the highest levels of The relationships within the NA genera group homoplasy (CI # 0.13) include primary ray are poorly resolved and similar to those length (No. 17; CI 5 0.10), length of calyx inferred in the proportional weighting ap- teeth in fruit (No. 32; CI 5 0.13), width of proach. lateral wings (No. 45; CI 5 0.13), ratio of wing length/width in cross-section (No. 50; CI 5 MORPHOLOGY:MAXIMUM PARSIMONY USING 0.11), presence/absence of strengthening cells GOLOBOFF CRITERION. MP analyses of mor- in fruits (No. 51; CI 5 0.08), and the presence/ phological data with Goloboff criterion select- absence of an accessory oil tube in the ribs ed resulted in the preset limit of 20,000 most (No. 54; CI 5 0.07). Overall, the homoplastic parsimonious trees, each of 549 steps (CI 5 characters have CI values ranging from 0.07 to 0.14; RI 5 0.62; RC 5 0.09; G-fit 5224.37). 0.87. The topology of the strict consensus tree (not shown) is very similar to that of the successive MORPHOLOGY:MAXIMUM PARSIMONY USING approximations approach, but slightly less EQUAL WEIGHTS. MP analyses of 54 morpho- resolved. logical characters, using equal weights, result- ed in the preset limit of 20,000 most parsimo- MORPHOLOGY:BAYESIAN. Among a total of nious trees, each of 491 steps (CI 5 0.16; RI 5 40,000 trees generated in the Bayesian analysis 0.67; RC 5 0.11; strict consensus tree not of 54 morphological characters, 10,000 trees shown). Again, resolution of relationships and were discarded as ‘‘burn-in’’ before the Ln BS support values are generally low, and over likelihood values stabilized. The remaining half of the branches (28 out of 54) occurring in 30,000 trees were used to generate a majority Fig. 1 are maintained. In this analysis, five rule consensus tree (not shown). The –Ln characters (vs. six in the proportionally values of these 30,000 trees ranged from weighted analysis) occur without homoplasy 2176.31 to 2296.60, averaged 2227.62, with a (Nos. 14, 20, 25, 39, and 46). Two previous standard deviation of 14.50. Relationships non-homoplastic characters are now homo- inferred by the Bayesian tree are very similar plastic (No. 21, CI 5 0.75; No. 22, CI 5 0.67). to, or consistent with, those estimated by One previous homoplastic character is now proportionally or equally weighted MP meth- non-homoplastic (No. 25, CI 5 1.00). Again, ods (Fig. 1). many characters emphasized in previous classification systems show high levels of MORPHOLOGY:PHYLOGENETIC RESOLUTIONS. homoplasy. Characters exhibiting the highest The results of each morphological analysis levels of homoplasy are the same as those showed that the resolution of relationships identified in the proportionally weighted MP among these NA taxa is low, with many clades analysis and have similar CI values. Overall, not very well-supported. Cymopterus (bold- the homoplastic characters have CI values faced in Figs. 1 and 2) and Lomatium,twoof ranging from 0.05 to 0.83. the largest genera, are highly polyphyletic, as are many other genera within the group. MORPHOLOGY:MAXIMUM PARSIMONY USING Constraining the 40 examined taxa of Cym- SUCCESSIVE APPROXIMATIONS. MP analyses of opterus to monophyly and rerunning the 54 morphological characters using successive equally or proportionally weighted MP anal- approximations starting with proportional or ysis of morphological characters resulted in equal weights each resulted in the preset limit trees of 11 or 9.5 steps longer than those most of 20,000 most parsimonious trees, each of parsimonious trees obtained without the 52.04 and 51.67 steps, respectively (CIs 5 0.38 constraint invoked (491 or 331.83 steps, and 0.38, RIs 5 0.81 and 0.82, and RCs 5 respectively). Lomatium arose as monophyletic 0.31 and 0.31, respectively; strict consensus in trees of 7 or 6.17 steps longer than those trees not shown). For both of these analyses, without the constraint. Similar analyses re- five iterations were needed to stabilize the RC vealed that Aletes, Musineon, Oreoxis, Podis- values, both from initially 0.11 to 0.31. Glehnia tera, Pseudocymopterus, Pteryxia,and 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 141

Tauschia are each monophyletic in trees 2 to 7 Angelica (BS value ,50%), and the third steps greater than those most parsimonious. branch comprising all other accessions of the Given the large number of steps required to perennial, endemic NA genera group. The force monophyly of most of these genera, it is latter comprises a highly branched polytomy. highly unlikely that they represent natural Greater resolution of ingroup relationships is groups. With few exceptions, none of the obtained when the morphological characters major clades or subclades revealed coincide are given proportional weights in the com- with traditionally recognized genera or infor- bined analysis. In the strict consensus tree mally recognized species groups based on derived from this analysis (Fig. 2), the mono- morphology. Only Oreonana, Orogenia, Gleh- phyly of the perennial, endemic NA genera nia, Polytaenia,andPodistera are each re- group continues to be supported, with Glehnia vealed as monophyletic, while Thaspium, being its basalmost lineage. The nine acces- Zizia, Oreoxis,andPseudocymopterus are sions of Angelica constitute a clade that is monophyletic only in some of the analyses. sister group to all aforementioned taxa. The monophyly of the outgroup Angelica is Constraining Cymopterus to monophyly and supported, but with the exclusion of A. rerunning the MP analysis, with morpholog- capitellata. This species is different from its ical characters given either equal or propor- congeners by having its umbellets covered by a tional weights, resulted in trees 44 and 54.33 woolly indumentum. Other characters distin- steps longer than those most parsimonious guishing this taxon from its congeners include obtained without the constraint invoked (2454 an irregularly toothed leaf margin, herbaceous and 2239.33 steps, respectively); Lomatium bractlets with narrow scarious margin, longer arose as monophyletic in trees 25 and 35.33 calyx teeth in fruit, thick and corky fruit steps longer. Similar analyses revealed that wings, and the absence of strengthening cells Aletes, Musineon, Oreoxis, Podistera, Pseudo- in the fruit. Previously, this striking species cymopterus, Pteryxia,andTauschia are each was recognized as Sphenosciadium capitellatum monophyletic in trees six to 38.83 steps greater A. Gray, but was subsequently transferred than those most parsimonious. into Angelica on the basis of molecular data (Spalik et al. 2004). Molecular and combined COMBINED MORPHOLOGICAL AND MOLECULAR morphological/molecular data (presented im- CHARACTERS:BAYESIAN. Among a total of mediately below) support the monophyly of all 20,000 trees generated in the Bayesian analy- Angelica species. sis, 5,000 trees were discarded as ‘‘burn-in’’ before the Ln likelihood values stabilized; COMBINED MORPHOLOGICAL AND MOLECULAR 15,000 of these trees were used to generate a CHARACTERS:MAXIMUM PARSIMONY. MP anal- majority rule consensus tree (not shown). The yses of combined (morphological and molec- –Ln values of these 15,000 trees ranged from ular) data, giving either proportional or equal 19578.08 to 19676.28, averaging 19618.10, weights to the morphological characters, with a standard deviation of 13.55. Relation- resulted in either 240 minimal length trees ships inferred by the Bayesian tree are (each of 2239.33 steps, CIs 5 0.44 and 0.30, identical to, or highly consistent with, those with and without uninformative characters; RI estimated by MP analysis with morphological 5 0.64; RC 5 0.28) or the preset limit of characters given proportional weights. Bayes- 20,000 minimal length trees (each of 2454 ian PP values are presented on the MP strict steps; CIs 5 0.41 and 0.28, with and without consensus tree (Fig. 2). uninformative characters; RI 5 0.62; RC 5 0.26), respectively. Less resolution was COMBINED MORPHOLOGICAL AND MOLECULAR achieved in the strict consensus tree derived CHARACTERS:PHYLOGENETIC RESOLUTIONS. The from equally weighted MP analysis of com- least number of most parsimonious trees is bined data. In this tree (not shown), the derived from the combined analysis when monophyly of the perennial, endemic NA morphological characters are given propor- genera group is weakly supported (BS value tional weights, and these trees are largely 52%). A basal trichotomy is identified, with congruent with those trees derived from the first branch containing the two subspecies molecular data alone (Sun 2003; Sun and of Glehnia littoralis (BS value 75%), the second Downie, 2010). In fact, these trees are better branch containing all nine accessions of resolved and, in general, their branches more 142 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137

FIG. 2. Strict consensus tree of 240 minimal length trees derived from MP analysis of combined molecular (ITS, rps16 intron, and trnF-L-T) and morphological (proportionally weighted) characters for 129 accessions of NA Apioideae. The four major clades of Apioideae circumscribed previously (Sun 2003; Sun and Downie, 2010) are indicated. Numbers on branches are bootstrap estimates (BS) for 100,000 replicate analyses using ‘‘fast’’ stepwise addition and Bayesian posterior probability (PP) values expressed as percentages, respectively; values ,50% for both support values are either not indicated or indicated by ‘‘--’’. 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 143 strongly supported than any phylogenetic tree species of Cymopterus (C. aboriginum, C. for the group heretofore available, thus we use basalticus, C. cinerarius, C. evertii, C. gilmanii, these results to tentatively circumscribe major C. globosus, C. lapidosus, C. ripleyi). Cymop- clades and subclades to facilitate future terus aboriginum is not allied with this group phylogenetic and revisionary studies of this on the Bayesian tree (not shown). These eight problematic group. For consistency with our species share dorsally compressed fruits and earlier studies, we identify the four major the absence of a carpophore (except C. clades circumscribed previously on the basis of aboriginum and C. lapidosus). Although most MP analysis of all available molecular data of these species were at one time recognized in (Clades 1–4, Fig. 2), even though Clade 2 the genus Aulospermum (Mathias 1930), there arises from within a paraphyletic Clade 1 in all is no unique and obvious morphological MP analyses of combined morphological and synapomorphy supporting this subclade. A molecular data presented herein. The results of putative close relationship among C. aborigi- the Bayesian analysis of these same combined num, C. cinerarius,andC. evertii was suggest- data, however, revealed Clades 1 and 2 as ed by Hartman and Kirkpatrick (1986). Three monophyletic sister groups. Here, we continue character state changes occurred along the to treat Clade 2 as separate from Clade 1, as branches leading to Subclade 1a: No. 12, previous molecular studies and the Bayesian changing from states 2 to 0; No. 15, changing analysis of combined data have revealed. The from states 0 to 1; and No. 41, changing from monophyletic genus Glehnia cannot be as- state 0 to equivocal; however, all of these signed to any of the three major ingroup character state changes have reversals within clades, thus further studies are warranted to the subclade. Subclade 1b contains four clarify its phylogenetic relationships. Of the species of Cymopterus (C. jonesii, C. minimus, four major clades identified, one represents the C. purpureus, C. rosei). These species were also outgroup genus Angelica and will not be circumscribed in Aulospermum (Mathias 1930) discussed further. Each of the other three and all are morphologically very similar. major clades contains three to eight subclades, Indeed, C. jonesii, C. minimus,andC. rosei several of which are moderately or well- were treated as varieties of C. purpureus supported in either the MP or Bayesian (Goodrich 1986). All species share the pres- analyses. Overall, while the combined analyses ence of a pseudoscape. Subclade 1c comprises confirmed the monophyly of the perennial, the five varieties of Cymopterus acaulis (i.e., endemic NA Apioideae, many clades and vars. acaulis, fendleri, greeleyorum, higginsii, subclades are weakly supported, with most of and parvus), with C. newberryi closely allied. these having no resemblance to pre-established Cymopterus newberryi has a similar leaf groups. Only five traditionally recognized morphology to that of C. acaulis, but it varies genera (Oreonana, Orogenia, Thaspium, Zizia, greatly in wing development (the latter has and Polytaenia) are revealed as monophyletic well-developed wings, whereas the former has in the combined analyses. For those species dorsal wings similar to the lateral or often with infraspecific taxa or those represented by narrower and irregularly developed, or they more than one accession, only three (Cymop- may even be obsolete, thus resembling the terus acaulis, Glehnia littoralis, Pseudocymop- situation in Lomatium). Based on their similar terus montanus) are revealed as monophyletic; habit, C. newberryi was treated as a variety of three other species (Aletes acaulis, Aletes C. fendleri (Jones 1908). This group is macdougalii, Pteryxia terebinthina) are para- supported by the presence of a pseudoscape, phyletic or polyphyletic. dorsally compressed fruits, and dorsal wings, Clade 1 contains 60 accessions, representing and the absence of a carpophore. Subclade 1d 10 genera. The genera Neoparrya, Oreonana, represents another six species of Cymopterus and Shoshonea occur exclusively in this clade. (C. corrugatus, C. coulteri, C. deserticola, C. Thirty accessions of Cymopterus (representing douglassii, C. ibapensis, C. nivalis). This group 75% of all accessions of Cymopterus included is paraphyletic on the Bayesian tree (not in this study) also occur here. Eight subclades shown), with Subclade 1e arising from within (1a-h), each containing three to eight acces- it. Cymopterus corrugatus and C. coulteri are sions, are recognized. Most of these subclades very similar morphologically, both having have BS values of 51–86% and PP values of wavy wings and ternate or pinnate leaves; on 90–100%. Subclade 1a is composed of eight the basis of this similarity, Jones (1908) treated 144 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137

C. coulteri as a variety of C. corrugatus. All carpum, L. orientale, L. triternatum subsp. species are acaulescent and possess irregularly platycarpum). This subclade is not revealed by toothed leaf margins. Subclade 1e contains the Bayesian analysis. The group is character- four accessions of Pteryxia, i.e., P. petraea, P. ized by dorsally compressed fruits without terebinthina var. albiflora (two accessions), and dorsal wings, features typical of Lomatium P. terebinthina var. calcarea. These taxa all species. bear dorsally compressed fruits with dorsal Clade 3 comprises 41 accessions. These wings. Pteryxia petraea has sometimes been represent 14 genera, with seven (i.e., Har- treated as a variety of the P. terebinthina bouria, Musineon, Polytaenia, Pseudocymop- complex (Goodrich 1986; Constance 1993). terus, Taenidia, Thaspium,andZizia)found Subclade 1f is composed of three species of exclusively in this clade. Six subclades are Oreonana (O. clementis, O. purpurascens, O. circumscribed, supported by BS values rang- vestita). The monophyly of this group is ing from less than 50% to 100% and mostly supported by the rigid and persistent pedicels high PP values (96–100%). Subclade 3a of its sterile flowers. Subclade 1g contains contains six species of the Phellopterus group three accessions of Aletes macdougalii subsp. (Coulter and Rose 1900; Mathias 1930; Hart- breviradiatus and one accession of Oreoxis man 2000) of Cymopterus (C. bulbosus, C. trotteri. These two taxa are considered as constancei, C. macrorhizus, C. montanus, C. being conspecific (S. Goodrich et al., unpub- multinervatus, C. purpurascens). These plants lished data). They share an extremely similar share large and showy bractlets that are often leaf morphology, laterally compressed fruits, basally connate. However, similar bractlets and the presence of one oil tube in each also occur in C. basalticus (Subclade 1a). The interval of their fruits. Aletes humilis is closely species of the Phellopterus group and C. allied to this group in both MP and Bayesian basalticus differ in their leaf morphology; the trees. Subclade 1h constitutes three varieties of latter has palmately dissected leaves with three Pteryxia terebinthina (i.e., vars. californica, overlapping leaflets, whereas those of the foeniculacea,andterebinthina). These taxa also former have pinnately and more openly share an extremely similar leaf morphology. dissected leaves. Subclade 3a is also supported Clade 2 comprises 16 accessions from three by homoplastic characters, such as the pres- genera, representing two accessions of Oro- ence of a pseudoscape, dorsally compressed genia, 12 accessions of Lomatium (60% of all fruits, and dorsal wings. Subclade 3b compris- Lomatium accessions examined), and two es all species of Polytaenia, Thaspium,and accessions of Cymopterus. Three subclades Zizia, and each of these genera is monophy- are designated within this clade. Subclade 2a letic. The species of Thaspium and Zizia are contains both species of Orogenia (O. fusifor- remarkably similar in appearance and this mis and O. linearifolia). The monophyly of group is supported by the unique synapomor- Orogenia is supported by two unique morpho- phy of a fibrous and fascicled root system. The logical synapomorphies: corky, rib-like pro- generic limits of Thaspium and Zizia have been jections on its fruit axes, and incurved lateral questioned (Ball 1979; Lindsey 1982; Cooper- wings. Subclade 2b is composed of two species rider 1985), but in this study they comprise of Cymopterus (C. longipes, C. planosus)and well-supported monophyletic sister groups. two varieties of Lomatium grayi (vars. depau- Three character state changes occur along peratum and grayi). Both C. longipes and C. the branches leading to Subclade 3b: No. 17, planosus were also recognized in Aulospermum changing from states 1 to equivocal; No. 52, by Mathias (1930), but differ from each other changing from states equivocal to 0; and in flower color. The two varieties of L. grayi No. 53, changing from states 1 to 0. Two of are very similar morphologically, although these character state changes (Nos. 52 and 53) var. grayi has more ultimate leaf segments have no reversals within the subclade. Sub- and relatively larger fruits than those of var. clade 3c comprises five accessions representing depauperatum. MacClade revealed only one four genera, representing Aletes macdougalii character state change along the branches subsp. macdougalii, Cymopterus beckii, Pseu- leading to Subclade 2b: No. 10, changing docymopterus montanus (two accessions), and from states equivocal to 0. Subclade 2c is Pteryxia davidsonii. These plants share a linear composed of six species of Lomatium (i.e., L. leaf morphology. Aletes macdougalii subsp. bradshawii, L. cous, L. juniperinum, L. macro- macdougalii, C. beckii,andP. davidsonii 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 145 resemble each other morphologically, and the incurved lateral wings. The group of Thaspium first two taxa have been suggested as conspe- + Zizia is supported by having fibrous and cific (Hartman 2006). MacClade revealed no fascicled roots (No. 4, state 2). The Phellop- morphological character state changes along terus group of Cymopterus (C. bulbosus, C. the branches leading to Subclade 3c. Subclade constancei, C. macrorhizus, C. montanus, C. 3d contains five species representing four multinervatus,andC. purpurascens) is support- genera (Podistera macounii, P. yukonensis, ed by having a bract (No. 19, state 0) and a Lomatium brandegei, Musineon lineare,and complete involucel (No. 28, state 0), but these Taenidia integerrima). The two species of two characters also occur elsewhere on the Podistera share a stylopodium. This subclade tree, such as in Podistera yukonensis, P. is not supported by the morphological analy- macounii, C. glaucus,andC. basalticus. The ses. Three character state changes occur along genera Thaspium, Zizia,andPolytaenia, while the branches leading to Subclade 3d: No. 9, each revealed as monophyletic in the com- changing from states 0 to 2; No. 20, changing bined analyses, are also supported by a suite of from states 2 to equivocal; and No. 50, homoplastic characters. None of the three changing from states 1 to equivocal. Among major ingroup clades, circumscribed previous- these character state changes, only one (No. 9) ly on the basis of molecular evidence and has no reversals within the subclade. Subclade recovered herein in the Bayesian analysis of 3e is composed of five accessions representing combined data, are supported by unique four genera [Cymopterus williamsii, Musineon morphological synapomorphies. Similarly, tenuifolium, Oreoxis humilis,andAletes acaulis many of the subclades circumscribed herein (two accessions)]. Cymopterus williamsii was on the basis of combined morphological and once indicated as possibly belonging to molecular data are not supported by unique Oreoxis (Hartman and Constance 1985). morphological synapomorphies either. Clade 1 MacClade revealed four character state chang- is supported by only molecular data, with no es along the branches leading to Subclade 3e: morphological character state changes identi- No. 6, changing from states 0 to 1; No. 16, fied. Three morphological character state changing from states 0 to 1; No. 32, changing changes occur on the branch leading to Clade from states equivocal to 0; and No. 54, 2, and three along the branch leading to Clade changing from states 1 to 0. Subclade 3f 3 (Fig. 3). Reversals, however, are apparent contains Aletes sessiliflorus, A. filifolius, Har- for each of these characters within these bouria trachypleura, Oreoxis bakeri,andPseu- clades. These three major clades are not easily docymopterus longiradiatus. Neither subclade delimited, or cannot be delimited whatsoever, 3e nor subclade 3f is supported by the on the basis of morphology. morphological analyses. Two character state The distribution of six morphological char- changes occurred along the branches leading acters (seven character states) widely used in to Subclade 3f: No. 6, changing from states 0 traditional treatments of the group is provided to 2; and No. 41, changing from states 0 to in Fig. 3. Like most other morphological equivocal. characters, these six characters are highly homoplastic, each arising or being lost multi- MORPHOLOGICAL CHARACTER OPTIMIZATIONS. ple times during the evolution of the group. To assess evolutionary patterns of individual Optimization of the character ‘‘presence/ab- morphological characters and their usefulness sence of a pseudoscape’’ (No. 7) revealed that in genus and major clade determinations, each it required 20 steps on each of the 240 MP of the 54 morphological characters were trees. There are at least 13 gains for state 0, the optimized onto all of the 240 trees inferred presence of this character. Three lineages by MP analysis of combined morphological characterized by a pseudoscape each contain and molecular data. The results revealed that four to six accessions of Cymopterus. One of only two traditionally recognized genera, these lineages is composed of the six species of Oreonana and Orogenia, are supported by the Phellopterus group (Subclade 3a); a second unique synapomorphies. As stated above, comprises the five varieties of Cymopterus Oreonana is supported by having rigid and acaulis (Subclade 1c); and the third consists of persistent pedicels on its sterile flowers, and four species of Cymopterus (Subclade 1b). The Orogenia is supported by having corky and remaining lineages contain one to three rib-like projections on its fruit axes and accessions each, representing Cymopterus (15 146 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137

FIG. 3. Distribution of seven morphological character states on a randomly selected minimal length tree derived from MP analysis of combined molecular (ITS, rps16 intron and trnF-L-T) and morphological (proportionally weighted) characters for 129 accessions of NA Apioideae. Mapped character states are indicated on the figure and are as follows: 1, a pseudoscape is present (No. 7, state 0); 2, a stylopodium is present (No. 30, state 1); 3, fruits are dorsally compressed (No. 35, state 0); 4, a carpophore is absent (No. 36, state 2); 5, dorsal wings are absent (No. 40, state 1); 6, one oil tube is present in each interval (No. 52, state 0); 7, dorsal wings are present (No. 40, state 0). 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 147 accessions), Lomatium (five accessions), and and dorsal ribs are winged (state 0), only Musineon (one accession). The ancestral con- lateral ribs are winged (state 1), and no ribs are dition is identified as the absence of a winged (state 2). State 0 is revealed as the pseudoscape (state 1). Optimization of the ancestral condition. There are eight and four character ‘‘presence/absence of a stylopodium gains for character states 1 and 2, respectively. in fruit’’ (No. 30) revealed that this character Three of the lineages with dorsal wings required four steps on one randomly chosen (Subclades 1c + 1e, 3a, and Angelica) contain tree. Across all 240 MP trees, it required either six to 10 accessions of Cymopterus or Pteryxia. three or four steps. The ancestral condition is One of the lineages without dorsal wings (state revealed as the presence of a stylopodium in 1) contains six accessions of Lomatium (Sub- fruits (state 1), for this character occurs in clade 2c). All other lineages contain one to two Angelica (nine accessions) and Aethusa. With- accessions. The monophyly of Thaspium (three in the perennial, endemic NA Apioideae accessions) is supported by the presence of group, only Podistera (four species in three both dorsal and lateral wings. Optimization of lineages) is characterized by having a promi- the character ‘‘number of oil tubes in the nent stylopodium, although the genus is interval of the fruit’’ (No. 52) indicated that polyphyletic. Optimization of the character this character required 25 steps on each of the ‘‘dorsally/laterally compressed or terete fruits’’ 240 MP trees. The ancestral condition is the (No. 35) indicated that this character required presence of one oil tube in each interval in the 21 steps on one arbitrarily selected tree, and fruits (state 0). There are at least 10 losses and 21–22 steps when all of the 240 MP trees were 9 gains for character state 0. Two of the considered. While the ancestral condition of lineages with a single oil tube in each interval this character is equivocal, there are at least 10 contain six accessions each: one lineage losses and seven gains for character state 0 contains all but one species of Angelica; the (dorsally compressed fruits), eight gains for other is composed of three monophyletic state 1 (laterally compressed fruits), and four genera: Polytaenia, Thaspium,andZizia.All gains for character state 2 (terete fruits). The remaining lineages characterized by one oil last state occurs in Cymopterus williamsii, C. tube in an interval are composed of one to five douglassii, Shoshonea,andThaspium. Five of accessions each. the lineages with dorsally compressed fruits contain six to 10 accessions (Subclades 1a, 1c + Discussion. The results of diverse analyses of 1e, 2c, 3a, and Angelica). Two of the lineages both morphological and combined morpho- with laterally compressed fruit contain five to logical and molecular data are in agreement six accessions. One is composed of C. davisii, with our earlier studies based exclusively on Oreonana (three accessions), and Tauschia molecular evidence in revealing that many NA parishii, the other contains Subclade 1g and Apioideae genera are not monophyletic. The Podistera eastwoodiae. Optimization of the two largest genera, Lomatium and Cymop- character ‘‘development of a carpophore’’ terus, are each highly polyphyletic, with their (No. 36, CI 5 0.43) showed that this character species inextricably linked with each other and required 19 steps on each of the 240 MP trees. those of Aletes, Oreoxis, Pseudocymopterus, This character contains three character states: Pteryxia, and several other smaller genera of a carpophore is persistent (state 0), a carpo- the region. The results of the combined phore is present but falling with the mericarp analysis when morphological characters are (state 1), and a carpophore is absent (state 2). given proportional weights offer the most State 0 is revealed as the ancestral condition. resolved and best supported trees heretofore There are at least 9 gains for character state 2. available for the group. These trees are also Two of the lineages without a carpophore each fully congruent with those derived from contain six accessions of Cymopterus (Sub- molecular data alone. Four major clades are clade 1c with C. newberryi included, and recognized, one of which represents the out- Subclade 1a excluding C. lapidosus and C. group taxon Angelica. Numerous subclades aboriginum). Optimization of the character are also revealed, although very few are ‘‘development of wings on fruit ribs’’ (No. 40) supported by uniquely occurring morpholog- revealed that this character required 30 steps ical synapomorphies and many are supported on each of the 240 MP trees. This character is poorly in the combined analyses. The Phellop- divided into three character states: both lateral terus group of Cymopterus (C. bulbosus, C. 148 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137 constancei, C. macrorhizus, C. montanus, C. very limited use in delimiting genera and multinervatus,andC. purpurascens) may very major clades. The six characters used widely well be the only previously identified species in previous classifications of the group are groups within the complex that is supported highly homoplastic, resulting in many different by molecular and morphological evidence. treatments for the group and difficulties in Therefore, until these subclades receive con- circumscribing taxa unambiguously. firmation through additional study, we do not In Cymopterus, Cronquist (1997) reported formally recognize new assemblages of taxa at that some species (C. acaulis, C. bulbosus, C. the present time. These clades and subclades purpurascens) have a pseudoscape (a scape-like are only provisionally recognized and, pending stalk of a leaf cluster that originates from the support from further studies, will be used as a root-crown) with a subterranean root crown, framework in future phylogenetic and revi- whereas other species (C. aboriginum, C. sionary studies of NA Apioideae. cinerarius, C. nivalis) have a taproot sur- The monophyly of the entire group of mounted by a branching, surficial caudex. perennial, endemic Apiaceae subfamily Apioi- However, some species do not fit completely deae of NA is supported by both morpholog- into either of these categories. As examples, C. ical and molecular analyses. The restricted megacephalus and C. ripleyi have a simple distribution of many of these plants to subterranean root crown, but lack a pseudo- elevated regions of similar habitat, their scape. Cymopterus duchesnensis has a taproot similar life history and overall general habit, capped by a surficial crown or more often by a and their overlapping patterns of morpholog- branched caudex. Several other species have a ical character variation suggested previously surficial or subterranean root crown, but do or that this group of umbellifers was closely do not have a pseudoscape. Our results show related. The absence of a prominent conical that the ancestral condition is the absence of a stylopodium in all taxa except Podistera, pseudoscape and that the derivation of a where the stylopodium is otherwise well pseudoscape has been achieved multiple times developed, as it is in most other umbellifers, during the evolution of the group. A pseudo- is a synapomorphy uniting the group. Further scape is also present in some species of support for their monophyly comes from the Lomatium (L. juniperinum, L. cous, L. macro- shared presence of a protogynous breeding carpum, L. bicolor)andMusineon (M. divar- system (and associated reproductive charac- icatum), thus its presence has limited utility for teristics), an atypical feature in a family where reliably delimiting taxa. Similarly, plants floral protandry prevails (Schlessman et al. having a taproot surmounted by a branching, 1990). Protogyny is presumed derived in the surficial caudex are also found in multiple apioid umbellifers, a response to an early separate lineages (not shown). flowering season and unreliable pollinators The presence of a prominent conical stylo- (Schlessman and Graceffa 2002). podium (a disc-like to long-tapering enlarge- Fruit and other morphological characters ment borne atop the ovary at the base of the traditionally have been used to delimit taxa styles) is commonly present in many species of within Apioideae. However, heretofore, these Apiaceae, therefore, the absence of a stylopo- characters have not been analyzed cladistically dium is considered as prime evidence support- across a wide spectrum of NA taxa. Thus, in ing the monophyly of perennial, endemic NA the absence of a phylogenetic estimate, pat- Apiaceae subfamily Apioideae (Mathias and terns in the evolution of these characters and Constance 1944–1945; Downie et al. 2002; Sun their utility in circumscribing monophyletic et al. 2004). Only the genus Podistera within groups could not be properly assessed. In this the ingroup possesses a stylopodium. Howev- study, we have determined that only two er, because this genus is not monophyletic in traditionally recognized genera, Oreonana the combined analyses, the presence of a and Orogenia, are supported by uniquely- stylopodium arises three times independently. occurring morphological character states. This character is readily observable, but its The genera Thaspium, Zizia,andPolytaenia presence does not unambiguously circum- are also each revealed as monophyletic in the scribe any one particular genus within the combined analyses, but none of them are ingroup, as previously considered. supported by morphological synapomorphies. In Apiaceae, fruit compression patterns In general, morphological characters have have been used to distinguish taxonomic 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 149 groups at various levels. In western NA deserticola, C. douglassii, C. longipes, C. Apioideae, dorsally compressed fruits are megacephalus, C. newberryi, C. ripleyi, C. present in many genera, such as Cymopterus, williamsii) makes this character unreliable to Lomatium, Orogenia, Pteryxia, Pseudocymop- separate Cymopterus from Lomatium. Similar terus, Polytaenia, Glehnia,andAngelica. While fruits to those of typical Lomatium are also definite laterally or dorsally compressed fruits seen in some species of Pteryxia (P. tere- are readily distinguishable in Cymopterus, binthina, P. hendersonii)andPseudocymop- there are numerous intermediate stages such terus (P. montanus). The presence of fruits that ‘‘the interpretation [of orientation of fruit with both lateral and dorsal wings supports compression] depends on the individual’s seven of the subclades designated herein, as point of view’’ (Mathias 1930). Fruit cross- well as several other taxa (such as, Thaspium, sections reveal a complex series, from fruits Glehnia,andAngelica). Two subclades and that are subterete to somewhat compressed Polytaenia are supported by the absence of laterally (C. douglassii, C. jonesii, C. longipes, dorsal wings. C. nivalis, C. panamintensis) to those that are The number of vittae (oil tubes) in the markedly compressed dorsally (C. deserticola, intervals between the primary ribs of the C. newberryi). Our results show that dorsally fruits was used to distinguish primarily compressed fruits support many separate between Aletes (mostly solitary) and Neopar- subclades, as do laterally compressed fruits. rya (numerous; Theobald et al. 1963). Cron- These results agree with those obtained in quist (1997) submerged Aletes into Musineon other studies of Apiaceae, where the orienta- because the distinction between some species tion of fruit compression, a feature used of Aletes and Musineon is no more than the widely in traditional systems of classification number of oil tubes (two or more in the of the family, is an unreliable character for latter). In Cymopterus, this number varies circumscribing taxa (Cronquist 1982; Downie from 3 to 5. All but one species of Angelica are et al. 2001). supported by the presence of one oil tube in Nearly half of the species of Cymopterus each fruit interval, as are Glehnia, Polytaenia, lack a carpophore, a remnant of the floral axis Thaspium,andZizia. Subclade 1g is the only to which the mericarps are attached (Hartman subclade designated herein having a single and Constance 1985; Cronquist 1997; Hart- interval vitta in all included taxa. The genus man 2000). Our results show that the loss of Aletes is polyphyletic in all trees, thus the the carpophore (through adnation of its halves presence of a solitary vitta in the intervals of to the mericarps) has been independently the fruit is a highly homoplastic character achieved several times within Cymopterus. (arising at least four times independently in The absence of a carpophore also occurs in nine accessions) and cannot be used by itself all or some accessions of Aletes, Thaspium, to delimit genera. Oreoxis, Pseudocymopterus, Shoshonea,and In conclusion, our study confirms that Orogenia, supporting the monophyly of Or- morphological characters are of limited value ogenia and Thaspium. The presence of a for delimiting most traditionally-defined gen- carpophore supports the monophyly of Or- era within the group of perennial, endemic NA eonana, Zizia, Angelica,andPolytaenia.In apioid umbellifers. Many of these genera are total, this character is lost at least nine times ill-formed, being based on highly homoplastic within the group, and only serves to distin- and overlapping characters. Thus, the empha- guish two of the subclades designated herein. sis placed on these characters in previous The outer surface of the mericarp normally systems of classification of the group has led has five primary ridges or ribs (three dorsal to highly artificial assemblages of taxa. Such a and two lateral), in which the dorsal and/or conclusion is not surprising, given the com- lateral ribs may develop into wings. In general, mon disagreement among systematists in using species of Cymopterus bear (one to three) these characters to circumscribe higher-level wings on their dorsal fruit ribs, whereas in taxa within the family (e.g., Heywood 1971; Lomatium, the dorsal ribs are generally Theobald 1971; Davis 1972; Cronquist 1982). filiform and wingless or occasionally very Indeed, the results of numerous molecular narrowly winged. However, the absence of systematic investigations provide very little (or obsolete) dorsal wings found in some support for all but a few suprageneric taxa species of Cymopterus (C. corrugatus, C. erected on the basis of anatomical and 150 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137 morphological features of the mature fruit CATALA´ N, M. P., E. A. KELLOGG, AND R. G. (summarized in Downie et al. 2001). Generic OLMSTEAD. 1997. Phylogeny of Poaceae subfam- ily Pooideae based on chloroplast ndhF gene delimitation in Apiaceae is often vague and sequences. Mol. Phylogenet. Evol. 8: 150–166. arbitrary (Constance 1987; Cronquist 1997), CONSTANCE, L. 1987. Neogeozia (Apiaceae), a very and many species-rich genera are polyphyletic distinct and elegant genus of Mexican Umbellif- (Downie et al. 2000a, 2000b; Spalik et al. erae. Opera Bot. 92: 59–71. 2001). Unfortunately, the results of this study CONSTANCE, L. 1993. Apiaceae [Umbelliferae]: carrot family, p. 136–166. In J. C. Hickman [ed.], The do little to refute these statements. Of all the Jepson Manual: higher plants of California. perennial, endemic apioid genera of NA, only University of California Press, Berkeley and Oreonana, Orogenia, Polytaenia, Thaspium, Los Angeles, CA. and Zizia are each resolved as monophyletic COOPERRIDER, T. S. 1985. Thaspium and Zizia on the basis of phylogenetic analyses of (Umbelliferae) in Ohio. Castanea 50: 116–119. COULTER,J.M.AND J. N. ROSE. 1900. Monograph combined molecular and morphological data. of the North American Umbelliferae. Contr. Furthermore, all but a few of the major clades U.S. Natl. Herb. 7: 1–256. and subclades circumscribed herein are sup- CRONQUIST, A. 1982. Reduction of Pseudotaenidia to ported by homoplastic morphological charac- Taenidia (Apiaceae). Brittonia 34: 365–367. ters. The systematics of the group is no where CRONQUIST, A. 1997. Apiaceae, p. 340–427. In A. Cronquist, N. H. Holmgren, and P. K. Holmgren near satisfactory, and a complete reassessment [eds.], Intermountain Flora: Vascular Plants of of the generic limits of these taxa is required. the Intermountain West, USA., Vol. 3, Part A. The systematic investigation of the perennial, The New York Botanical Garden, Bronx, NY. endemic genera of NA Apioideae needs to be DAVIS, P. H. 1972. Umbelliferae, p. 265–538. In continued with the goal of uncovering mor- Flora of Turkey and the East Aegean Islands, Vol. 4. University Press, Edinburgh. phological synapomorphies useful for clade DOWNIE, S. R., R. L. HARTMAN, F.-J. SUN, AND D. S. determination. If such synapomorphies cannot KATZ-DOWNIE. 2002. Polyphyly of the spring- be identified, we would have to accept that the parsleys (Cymopterus): Molecular and morpho- task of reclassifying this group is to be logical evidence suggest complex relationships accomplished on the basis of molecular among the perennial, endemic genera of western North American Apiaceae. Can. J. Bot. 80: evidence rather than on morphological data. 1295–1324. If future studies support the conclusions DOWNIE, S. R., D. S. KATZ-DOWNIE, AND M. F. presented herein, and if further resolution of WATSON. 2000a. A phylogeny of the flowering relationships can be achieved, radical changes plant family Apiaceae based on chloroplast to the prevailing classification of the perennial, DNA rpl16 and rpoC1 intron sequences: To- wards a suprageneric classification of subfamily endemic NA Apioideae will be required. Apioideae. Am. J. Bot. 87: 273–292. Indeed, such changes appear to be underway DOWNIE, S. R., G. M. PLUNKETT,M.F.WATSON,K. already. In accordance with previous floristic SPALIK,D.S.KATZ-DOWNIE,C.M.VALIEJO- studies (Goodrich 1986; Cronquist 1997), the ROMAN,E.I.TERENTIEVA,A.V.TROITSKY,B.-Y. genera Aletes (in part), Oreoxis, Pseudocymop- LEE,J.LAHHAM, AND A. EL-OQLAH. 2001. Tribes and clades within Apiaceae subfamily Apioideae: terus,andPteryxia have been recently includ- The contribution of molecular data. Edinb. J. ed within a broadly defined Cymopterus in a Bot. 58: 301–330. flora of the San Juan Basin region (S. Good- DOWNIE, S. R., S. RAMANATH,D.S.KATZ-DOWNIE, rich et al., unpublished data). The distinction AND E. LLANAS. 1998. Molecular systematics of between Lomatium and Cymopterus also Apiaceae subfamily Apioideae: phylogenetic analyses of nuclear ribosomal DNA internal remains very unclear, with no obvious char- transcribed spacer and plastid rpoC1 intron acter consistently separating these taxa. Given sequences. Am. J. Bot. 85: 563–591. this trend and overlapping character variation DOWNIE, S. R., M. F. WATSON,K.SPALIK, AND D. S. among genera, it may very well be possible KATZ-DOWNIE. 2000b. Molecular systematics of Old World Apioideae (Apiaceae): relationships that future studies will indicate that all or most among some members of tribe Peucedaneae members of the group should be combined sensu lato, the placement of several island- into one large, polymorphic genus, an extreme endemic species, and resolution within the apioid but possibly inevitable action. superclade. Can. J. Bot. 78: 506–528. FARRIS, J. S. 1969. A successive approximations Literature Cited approach to character weighting. Syst. Zool. 18: 374–385. BALL, P. W. 1979. Thaspium trifoliatum (meadow- FELSENSTEIN, J. 1985. Confidence limits on phylog- parsnip) in Canada. Can. Field Naturalist 93: enies: an approach using the bootstrap. Evolu- 306–307. tion 39: 783–791. 2010] SUN AND DOWNIE: NORTH AMERICAN APIACEAE SUBFAMILY APIOIDEAE 151

GILMARTIN,A.J.AND K. S. SIMMONS. 1987. flora, Vol. 28B. The New York Botanical Relationships of Lomatium among other genera Garden, Bronx, NY. of Apiaceae. Plant Syst. Evol. 157: 95–103. NIXON,K.C.AND J. M. CARPENTER. 1996. On GOLOBOFF, P. A. 1993. Estimating character weights simultaneous analysis. Cladistics 12: 221–241. during tree search. Cladistics 9: 83–91. RONQUIST,F.AND J. P. HUELSENBECK. 2003. MrBayes GOODRICH, S. 1986. Utah flora: Apiaceae (Umbel- 3: Bayesian phylogenetic inference under mixed liferae). Great Basin Naturalist 46: 66–106. models. Bioinformatics 19: 1572–1574. HARTMAN, R. L. 1985. A new species of Cymopterus SCHLESSMAN,M.A.AND L. M. GRACEFFA. 2002. Pro- (Umbelliferae) from southern Idaho. Brittonia togyny, pollination, and sex expression of andro- 37: 102–105. monoecious Pseudocymopterus montanus (Apia- HARTMAN, R. L. 1986. Studies in Rocky Mountain ceae, Apioideae). Int. J. Plant Sci. 163: 409–417. species of Cymopterus (Apiaceae). Am. J. Bot. SCHLESSMAN, M. A., D. G. LLOYD, AND P. P. LOWRY 73: 765–766 (Abstract). II. 1990. Evolution of sexual systems in New HARTMAN, R. L. 2000. A new species of Cymopterus Caledonian Apiaceae. Mem. N. Y. Bot. Gard. (Apiaceae) from the Rocky Mountain region, U. 55: 105–117. S. A. Brittonia 52: 136–141. SPALIK, K., J.-P. REDURON, AND S. R. DOWNIE.2004. HARTMAN, R. L. 2006. New combinations in the The phylogenetic position of Peucedanum sensu genus Cymopterus (Apiaceae) of the southwest- lato and allied genera and their placement in ern United States. SIDA 22: 955–957. tribe Selineae (Apiaceae, subfamily Apioideae). HARTMAN,R.L.AND L. CONSTANCE. 1985. Two new Plant Syst. Evol. 243: 189–210. species of Cymopterus (Umbelliferae) from west- SPALIK, K., A. WOJEWO´ DZKA, AND S. R. DOWNIE. ern North America. Brittonia 37: 88–95. 2001. Delimitation of genera in Apiaceae with HARTMAN,R.L.AND L. CONSTANCE. 1988. A new examples from Scandiceae subtribe Scandicinae. Lomatium (Apiaceae) from the Sierran crest of Edinb. J. Bot. 58: 331–346. California. Madron˜o 35: 121–125. STEVENS, P. F. 1991. Character states, morphological HARTMAN,R.L.AND R. S. KIRKPATRICK. 1986. A variation, and phylogenetic analysis: A review. new species of Cymopterus (Umbelliferae) from Syst. Bot. 16: 553–583. northwestern Wyoming. Brittonia 38: 420–426. SUN, F.-J. 2003. A phylogenetic study of Cymopterus HEYWOOD, V. H. 1971. Systematic survey of Old and related genera (Apiaceae). Ph.D. disserta- World Umbelliferae, p. 31–42. In V. H. Hey- tion–University of Illinois at Urbana-Cham- wood [ed.], The Biology and Chemistry of the paign, Urbana, IL. Umbelliferae. Academic Press, New York. SUN, F.-J. AND S. R. DOWNIE. 2004. A molecular HILLIS,D.M.AND J. P. HUELSENBECK. 1992. Signal, systematic investigation of Cymopterus and its noise, and reliability in molecular phylogenetic allies (Apiaceae) based on phylogenetic analyses analyses. J. Hered. 83: 189–195. of nuclear (ITS) and plastid (rps16 intron) DNA JONES, M. E. 1908. New species and notes. Contr. sequences. S. African J. Bot. 70: 407–416. West. Bot. 12: 1–81. SUN, F.-J. AND S. R. DOWNIE. 2010. Phylogenetic KAGAN, J. S. 1986. A new species of Lomatium relationships among the perennial, endemic (Apiaceae) from southwestern Oregon. Madron˜o Apiaceae subfamily Apioideae of western North 33: 71–75. America: additional data from the cpDNA trnF- KARTESZ, J. T. 1994. A synonymized checklist of the trnL-trnT region continue to support a highly vascular flora of the United States, Canada, and polyphyletic Cymopterus. Botanische Jahrbu¨- Greenland, Second edition. Biota of the North cher, in press. American Program of the North Carolina SUN, F.-J., S. R. DOWNIE, AND R. L. HARTMAN.2004. Botanical Garden. Timber Press, Portland, OR. An ITS-based phylogenetic analysis of the KLUGE, A. G. 1989. A concern for evidence and a perennial, endemic Apiaceae subfamily Apioi- phylogenetic hypothesis of relationships among deae of western North America. Syst. Bot. 29: Epicrates (Boidae, Serpentes). Syst. Zool. 38: 419–431. 7–25. SUN,F.-J.,G.A.LEVIN, AND S. R. DOWNIE. 2005. A KLUGE,A.G.AND A. J. WOLF. 1993. Cladistics: multivariate analysis of Cymopterus glomeratus, What’s in a word? Cladistics 9: 183–199. formerly known as C. acaulis (Apiaceae). Rho- LINDSEY, A. H. 1982. Floral phenology patterns and dora 107: 359–385. breeding systems in Thaspium and Zizia (Apia- SUN,F.-J.,G.A.LEVIN, AND S. R. DOWNIE. 2006. A ceae). Syst. Bot. 7: 1–12. multivariate analysis of Pseudocymopterus (Apia- MADDISON,D.R.AND W. P. MADDISON. 2003. ceae). J. Torrey Bot. Soc. 133: 499–512. MacClade 4: Analysis of phylogeny and charac- SUN,F.-J.,G.A.LEVIN, AND S. R. DOWNIE. 2008. A ter evolution, version 4.06. Sinauer Associates, multivariate analysis of Pteryxia terebinthina Sunderland, MA. (Apiaceae). J. Torrey Bot. Soc. 135: 81–93. MATHIAS, M. E. 1930. Studies in the Umbelliferae. SWOFFORD, D. L. 2003. PAUP*. Phylogenetic III. A monograph of Cymopterus including a Analysis Using Parsimony (*and other methods), critical study of related genera. Ann. Missouri version 4 edition. Sinauer Associates, Sunder- Bot. Gard. 17: 213–476. land, MA. MATHIAS, M. E. 1965. Distribution patterns of THEOBALD, W. L. 1971. Comparative anatomical certain Umbelliferae. Ann. Missouri Bot. Gard. and developmental studies in the Umbelliferae, 52: 387–398. p. 177–197. In V. H. Heywood [ed.], The Biology MATHIAS,M.E.AND L. CONSTANCE. 1944–1945. and Chemistry of the Umbelliferae. Academic Umbelliferae, p. 43–295. In North American Press, New York. 152 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL.137

THEOBALD, W. L., C. C. TSENG, AND M. E. MATHIAS. TORREY,J.AND A. GRAY. 1840. Umbelliferae, 1963. A revision of Aletes and Neoparrya p. 598–645. In A Flora of North America. Wiley (Umbelliferae). Brittonia 16: 296–315. & Putnam, New York. 2010] Appendix. Data matrix of 54 morphological characters used in the cladisti c analyses of 123 taxa of NA Apioideae. The symbol ‘‘?’’ indicates an inappl icable character state. Character states of polymorphic characters are indicat ed as letters: A 5 0, 1, or 2; B 5 0, 1, or 3; C 5 0, 2, or 3; D 5 0or1;E 5 0or2;F 5 0or3;G 5

1or2;H 5 1 or 3; and I 5 2or3. APIOIDEAE SUBFAMILY APIACEAE AMERICAN NORTH DOWNIE: AND SUN

Morphological characters 10 20 30 40 50 Taxon N N N N N Aethusa cynapium L. 1100101101 00G100101? ??00010101 0110201110 ?111210010 0001 Aletes acaulis (Torr.) J.M. Coult. & Rose 01E0011200 021101101? ??00010100 0010101110 ?111211010 0000 Aletes anisatus (A. Gray) W.L. Theob. & C.C. Tseng 0100001100 021100101? ??00010100 0010101110 ?111211010 0001 Aletes calcicola Mathias & Constance D1E0121100 031102101? ??00010100 1110101112 1????120?? 0110 Aletes filifolius Mathias, Constance & W.L. Theob. D100021120 121100101? ??00010100 0010101110 ?111211010 0001 Aletes humilis J.M. Coult. & Rose 01E0011200 02G100001? ??00010100 0010101112 0????110?? 0001 Aletes macdougalii J.M. Coult. & Rose subsp. macdougalii 01E0001200 021100001? ??00010100 0010101110 ?111211010 0001 Aletes macdougalii subsp. breviradiatus W.L. Theob. & C.C. Tseng 01E0001200 021100001? ??00010100 0010101110 ?1?1211010 0001 Aletes sessiliflorus W.L. Tseng & C.C. Tseng 0100001220 021100101? ??00010100 0000101112 1????110?? 0000 Angelica ampla A. Nelson 1100121111 00G102011? ??00010101 0110001110 ?111210010 0110 Angelica archangelica L. subsp. archangelica 1100121111 0CG101011? ??00010101 0110001110 ?111210010 0001 Angelica arguta Nutt. ex Torr. & A. Gray 11E0101111 0D2102011? ??1?????01 0110001110 ?111210011 0001 Angelica breweri A. Gray 11D0111111 002101111? ??1?????11 0111001110 ?111210011 0001 Angelica capitellata (A. Gray) Spalik, Reduron & S.R. Downie 1100101101 0D1101001? ??00011111 0101001110 ?011210010 1001 Angelica grayi (J.M.Coult.&Rose)J.M.Coult.& Rose 1120121111 011102111? ??000DG101 0110001110 ?111210010 0001 Angelica pinnata S. Watson 11E0101111 0D2102101? ??1?????21 0110001110 ?111210010 0011 Angelica roseana L.F. Hend. 11E0101111 0D2102111? ??00010121 0112001110 ?111210000 0011 Angelica sylvestris L. 1100121111 0DG101011? ??00010101 0110001110 ?111110011 0001 Cymopterus aboriginum M.E. Jones 0120001100 00011010D? ??00012100 01100D1110 ?011112111 0111 Cymopterus acaulis (Pursh) Raf. var. acaulis D100100100 002110101? ??0A01D100 010002?110 ?100110021 0110 Cymopterus acaulis var. fendleri (A. Gray) S. Goodrich D100100100 021110101? ??0A01D100 010002?110 ?100010021 0110 Cymopterus acaulis var. greeleyorum J.W. Grimes & P.L. Packard D100100100 021110101? ??0A01D100 010002?110 ?100010011 1110 Cymopterus acaulis var. higginsii (S.L. Welsh) S. Goodrich D100100100 010110101? ??0A010100 010002?110 ?100210011 0110 Cymopterus acaulis var. parvus S. Goodrich D100100100 021110101? ??0A01D100 010002?110 ?100210011 0110 Cymopterus basalticus M.E. Jones D100101000 0G1100101? ??00002000 010002?110 ?11D210010 1111 Cymopterus beckii S.L. Welsh & S. Goodrich 1100001E20 121110101? ??0001D100 0110110110 ?111211010 0001 Cymopterus bulbosus A. Nelson D100100100 0DE1100001 1001002000 0110001110 ?0DD110001 0110

Cymopterus cinerarius A. Gray 01G0001100 01112???1? ??00?1?100 010002?110 ?000010101 0111 153 Cymopterus constancei R.L. Hartm. D100100100 0D01101001 210G102000 0000001110 ?011010001 0110 Cymopterus corrugatus M.E. Jones D100100200 000100001? ??00012100 011002?110 ?100210020 1001 Cymopterus coulteri (M.E. Jones) Mathias D100100200 000120101? ??00012100 01D002?110 ?100110100 1001 154 Appendix. Continued.

Morphological characters 10 20 30 40 50 Taxon N N N N N Cymopterus davisii R.L. Hartm. D100001200 021110001? ??00011120 0112101110 ?111212110 1111 Cymopterus deserticola Brandegee 0100101100 01E12???1? ??1?????10 010102?111 00?1210020 1111 Cymopterus douglassii R.L. Hartm. & Constance 0100001200 021110001? ??0E011100 0010201111 11?1210010 1111 Cymopterus duchesnensis M.E. Jones D100D00200 021100101? ??00010100 0110001110 ?0DD010011 0110 [V SOCIETY BOTANICAL TORREY THE OF JOURNAL Cymopterus evertii R.L. Hartm. & R.S. Kirkp. 0120021100 0EA112101? ??000D1110 001202?111 11?1210010 0111 Cymopterus gilmanii C. Morton D100100000 0D0110001? ??0001D100 011002?110 ?111110011 0111 Cymopterus glaucus Nutt. D100100100 02G1001001 0100011100 0110001110 ?011212110 0111 Cymopterus globosus (S. Watson) S. Watson D10010D100 0D212???1? ??00002100 010002?110 ?111210020 0000 Cymopterus goodrichii S.L. Welsh & Neese 1100000100 0DG100001? ??0D011100 0110001110 ?000212110 0111 Cymopterus ibapensis M.E. Jones D1E0100100 000110001? ??0001D100 0110001110 ?011012111 0001 Cymopterus jonesii J.M. Coult. & Rose D100D00100 011100101? ??00010100 0110101110 ?0DD012101 0001 Cymopterus lapidosus (M.E. Jones) M.E. Jones D100100100 001110001? ??00010100 011000111D ?011210011 0110 Cymopterus longilobus (Rydb.) W.A. Weber 0100001100 021100101? ??00010100 10D0001110 ?011210011 1111 Cymopterus longipes S. Watson D100100100 0EG110101? ??00010100 011010111D 0011012111 0111 Cymopterus macrorhizus Buckley D100120100 0D010E1001 0101002000 0110001110 ?000110001 0110 Cymopterus minimus (Mathias) Mathias D1G010D100 0D2110101? ??00010100 01D0001110 ?011110001 0111 Cymopterus montanus Nutt. ex Torr. & A. Gray D1D0120100 0D011E0001 1101002000 0110001110 ?000110001 0110 Cymopterus multinervatus (J.M. Coult. & Rose) Tidestr. D100100100 0DE1200001 100111D000 01D002?110 ?0DD010001 0110 Cymopterus newberryi (S. Watson) M.E. Jones D120121200 021100101? ??00010100 01D002?11D ?101210020 0110 Cymopterus nivalis S. Watson 0100001100 0D0110001? ??00011100 0110101110 ?011210000 0111 Cymopterus panamintensis J.M. Coult. & Rose var. acutifolius (J.M. Coult. & Rose) Munz 0100001100 0I1100001? ??00010100 0110101110 ?011112011 1111 Cymopterus planosus (Osterh.) Mathias D100100100 0H2102101? ??00010100 0110001110 ?111212110 0111 Cymopterus purpurascens (A. Gray) M.E. Jones D100100100 0D01200001 1001102000 01D002?110 ?0DD010001 0110 Cymopterus purpureus S. Watson D1E0D0D100 0G1100101? ??00011100 01100D1110 ?0DD012101 0110 Cymopterus ripleyi Barneby 0100101000 0BG12???1? ??00002110 010102?111 1111210020 1111 Cymopterus rosei (M.E. Jones ex J.M. Coult. & Rose) M.E. Jones D1E0D2D200 0G1102101? ??00010100 0110001110 ?0DD110001 0111 Cymopterus williamsii R.L. Hartm. & Constance 0100001220 021110001? ??00011100 011022?111 11?1210010 0001 Glehnia littoralis F. Schmidt ex Miq. subsp. littoralis D1D0111201 000111101? ??00010110 0101001110 ?111110011 0010 Glehnia littoralis subsp. leiocarpa (Mathias) Hulte ´n D110111201 000111101? ??00010100 0100001110 ?111110011 0010 Harbouria trachypleura (A. Gray) J.M. Coult. & Rose 1100021120 1211D2001? ??00011120 0112101012 1????120?? 0001 Lomatium ambiguum (Nutt.) J.M. Coult. & Rose 1101101121 121100101? ??1?????00 0110001111 00?1210010 0001 Lomatium bicolor (S. Watson) J.M. Coult. & Rose var.

bicolor D101100100 021102101? ??00011100 0110001111 00?1210010 0001 OL

Lomatium bradshawii (Rose ex Mathias) Mathias & .137 Constance D100101121 021100101? ??02011100 0110001111 01?1210020 0221 Lomatium brandegei (J.M. Coult. & Rose) J.F. Macbr. 1100101120 12G100101? ??00011100 0110001111 00?1210010 0111 Lomatium californicum (Nutt.) Mathias & Constance 1100101101 021100001? ??1?????00 0110001111 01?1210010 0111 2010] Appendix. Continued.

Morphological characters APIOIDEAE SUBFAMILY APIACEAE AMERICAN NORTH DOWNIE: AND SUN 10 20 30 40 50 Taxon N N N N N Lomatium cous (S. Watson) J.M. Coult. & Rose D10112D101 021102101? ??01011110 0112001111 00?1210011 0001 Lomatium dasycarpum (Torr. & A. Gray) J.M. Rose subsp. dasycarpum D1D01D1101 0H110D101? ??00011110 011D001111 00?1210010 0DD0 Lomatium foeniculaceum (Nutt.) J.M. Coult. & Rose subsp. foeniculaceum 01D001D101 021101001? ??00D11110 0111001111 00?1210010 0110 Lomatium graveolens (S. Watson) Dorn & R.L. Hartm. var. graveolens 0100001120 121100001? ??00011100 0110001111 00?1210010 0111 Lomatium grayi (J.M.Coult.&Rose)J.M.Coult.& Rose var. grayi D100001100 021100101? ??00011100 0110001111 00?1210011 0001 Lomatium grayi var. depauperatum (M.E. Jones) Mathias D100001100 021100101? ??00011100 0110001111 00?1210011 0001 Lomatium junceum Barneby & N.H. Holmgren 0100001221 121100101? ??00010100 0110001111 00?1210011 0111 Lomatium juniperinum (M.E. Jones) J.M. Coult. & Rose D110110101 0EA101101? ??000121E0 0110001111 00?1210010 0111 Lomatium latilobum (Rydb.) Mathias 0100001220 021100001? ??00011100 0010001111 01?1210010 0DD1 Lomatium macrocarpum (Nutt. ex Torr. & A. Gray) J.M. Coult. & Rose 11DD11D101 0A2101101? ??000101D0 011D001111 00?1210011 0011 Lomatium nudicaule (Pursh) J.M. Coult. & Rose 0100101101 021100101? ??1?????00 0110001111 00?1210010 0001 Lomatium nuttallii (A. Gray) J.F. Machr. 0100001100 121100101? ??00011100 0010001111 00?1210010 0111 Lomatium orientale J.M. Coult. & Rose 1110011101 002101101? ??0D011100 0110001111 00?1210010 0111 Lomatium scabrum (J.M. Coult. & Rose) Mathias var. scabrum 01200D1100 021100101? ??00010100 0110001111 00?1210010 0111 Lomatium triternatum (Pursh) J.M. Coult. & Rose subsp. platycarpum (Torr.) Cronquist D110111121 121101101? ??00011110 0110001111 00?1210011 0001 Musineon divaricatum (Pursh) Nutt. ex Torr. & A. Gray var. divaricatum 11000E0G00 02111E001? ??00011100 0112101012 0????121?? 0110 Musineon lineare (Rydb.) Mathias D100001220 1E2112001? ??00010120 0112101012 0????120?? 1111 Musineon tenuifolium Nutt. ex Torr. & A. Gray 0100021G20 1E1112001? ??00011120 011210D012 0????120?? 0110 Musineon vaginatum Rydb. 1100021121 1E1112101? ??000101E0 0112100012 0????120?? 1111 Neoparrya lithophila Mathias 0100001220 021100101? ??00010100 0010101112 0????120?? 0110 Oreonana clementis (M.E. Jones) Jeps. 0010111100 000021001? ??02011120 0001101112 0????121?? 1111 Oreonana purpurascens Shevock & Constance 0010101200 000021011? ??02012120 0001101112 0????121?? 1111 Oreonana vestita (S. Watson) Jeps. 0010111100 0G1021001? ??00010120 0101101112 0????121?? 1111 Oreoxis alpina (A. Gray) J.M. Coult. & Rose subsp. alpina 0020011200 021111101? ??00010110 000112?110 ?111210010 0011

Oreoxis bakeri J.M. Coult. & Rose 0010021100 0E1112101? ??02010100 010012?110 ?111210010 0110 155 Oreoxis humilis Raf. 0000021200 021112001? ??00010100 000012?110 ?111110010 0010 Oreoxis trotteri S.L. Welsh & S. Goodrich 0020021200 021112001? ??00010100 000012?110 ?111210010 0001 Orogenia fusiformis S. Watson 0100101D20 100110101? ??1?????00 010002?101 0111200010 0111 Orogenia linearifolia S. Watson 0101101D20 100110101? ??1?????00 011002?101 0111200010 0111 Appendix. Continued. 156 Morphological characters 10 20 30 40 50 Taxon N N N N N Podistera eastwoodiae (J.M. Coult. & Rose) Mathias & Constance 0100001200 0I1110001? ??02010101 0010101012 0????120?? 0111 Podistera macounii (J.M. Coult. & Rose) Mathias & Constance 0100021220 01G1100000 E100010101 0010101012 0????120?? 0111 Podistera nevadensis (A. Gray) S. Watson 0020021220 02G11???1? ??01010101 0000101012 0????120?? 0111 Podistera yukonensis Mathias & Constance 0120021220 0EG1100000 0200010101 0010101012 0????120?? 0300 [V SOCIETY BOTANICAL TORREY THE OF JOURNAL Polytaenia nuttallii DC. 1120?21101 021102001? ??00010100 0110001111 0111210011 0000 Polytaenia texana (J.M. Coult. & Rose) Mathias & Constance 1120?21101 021102001? ??00010100 0110001111 0111210011 0000 Pseudocymopterus longiradiatus Mathias, Constance & W.L. Theob. D100021101 021102101? ??00010100 0110001110 ?D11210010 0001 Pseudocymopterus montanus (A. Gray) J.M. Coult. & Rose D1A0021GED DG1102D01? ??0D010100 01D00E111D 0011210011 0DD0 Pteryxia davidsonii (J.M. Coult. & Rose) Mathias & Constance 1100021120 0GG112101? ??00010110 0010001110 ?011211010 0111 Pteryxia hendersonii (J.M. Coult. & Rose) Mathias & Constance 0100001100 021100101? ??0E010100 10D0001110 ?011210011 0111 Pteryxia petraea (M.E. Jones) J.M. Coult. & Rose 1100001100 021100101? ??00010100 0010001110 ?011110011 0110 Pteryxia terebinthina (Hook.) J.M. Coult. & Rose var. terebinthina D100001101 021100101? ??0E010100 0010001110 ?000010011 0110 Pteryxia terebinthina var. albiflora (Nutt. ex Torr. & A. Gray) Mathias D100001101 002110101? ??0E010100 0010001110 ?011210011 0110 Pteryxia terebinthina var. calcarea (M.E. Jones) Mathias D100001101 02G110101? ??0E010100 0010001110 ?011110011 0110 Pteryxia terebinthina var. californica (J.M. Coult. & Rose) Mathias 1100001101 021100101? ??0E010100 0010001110 ?011110011 1110 Pteryxia terebinthina var. foeniculacea (Nutt. ex Torr. & A. Gray) Mathias D100001101 021110101? ??0E010100 0010001110 ?011210011 0110 Shoshonea pulvinata Evert & Constance 0020021220 021112101? ??00010120 00022G1112 1????100?? 0110 Taenidia integerrima (L.) Drude 1100101121 021100101? ??1?????00 1110101112 0????120?? 1111 Tauschia arguta (Torr. & A. Gray) J.F. Macbr. 1100101210 021100101? ??00010100 0110101112 0????121?? 1111 Tauschia glauca (J.M. Coult. & Rose) Mathias & Constance 1100101100 0G1100101? ??00010100 0100101112 0????121?? 1111 Tauschia kelloggii (A. Gray) J.F. Macbr. 0120101100 021100101? ??00010100 0110101112 0????121?? 1111 Tauschia parishii (J.M. Coult. & Rose) J.F. Macbr. 0100101100 021100001? ??00010100 0100101112 0????121?? 1111 Tauschia texana A. Gray 0100121200 021100101? ??00010100 0100101112 0????121?? 1111 Thaspium barbinode (Michx.) Nutt. 11D2121100 021102001? ??000111D0 111022?110 ?011212010 0001 Thaspium pinnatifidum (Buckley) A. Gray 1112121100 021102001? ??00011100 101222?110 ?011212010 0001 OL Thaspium trifoliatum (L.) A. Gray 1102121E10 01G102101? ??00011100 111022?110 ?011212010 0000 .137 Zizia aptera (A. Gray) Fernald 1102121E10 021102101? ??00011100 1110101112 0????120?? 0000 Zizia aurea (L.) W.D.J. Koch 1102121G10 021102101? ??00011100 1110101112 0????120?? 0000